Google is betting that the next great workplace productivity tool is not a new app but the browser window already open on every employee\u2019s screen. At Google Cloud Next on Tuesday, the company unveiled Auto Browse, a Gemini-powered feature that transforms Chrome from a passive web viewer into an agentic AI assistant capable of executing multi-step tasks on behalf of enterprise workers.

The announcement marks Google\u2019s most aggressive move yet to reposition Chrome\u2014already used by 3.8 billion people worldwide and commanding 67.7 percent of the global browser market\u2014as what the company is calling an \u201cintelligent workplace platform.\u201d

What Auto Browse Actually Does

Auto Browse, built on Google\u2019s latest Gemini 3 model, allows employees to delegate browser-based tasks through natural language commands. Rather than clicking through dozens of tabs and forms manually, workers can ask Gemini to handle multi-step workflows: researching hotel and flight costs across multiple date ranges, filling out expense reports, collecting tax documents, extracting competitor pricing data, or populating CRM fields from information in Google Docs.

\u201cThe new Gemini in Chrome is like having an assistant that helps you find information and get things done on the web easier than ever before,\u201d said Parisa Tabriz, VP of Chrome at Google, describing the shift from a browsing tool to \u201ca partner that can actually perform tasks on your behalf.\u201d

The feature leverages what Google calls the \u201cunique power of the browser\u201d to understand the live context of open tabs, enabling Gemini to work across multiple websites simultaneously. It also supports multimodal capabilities, allowing it to identify items in photos, search for similar products, and even apply discount codes while staying within budget parameters set by the user.

Human in the Loop, by Design

Google appears keenly aware that handing an AI agent the keys to an employee\u2019s browser carries significant risk. The company built what it describes as a \u201cdouble-check safety system\u201d that independently reviews the AI\u2019s planned actions before executing them.

\u201cAuto browse is designed to pause and explicitly ask for your confirmation or prompt you to complete some tasks like making a purchase or posting on social media,\u201d Tabriz explained. Strict boundaries limit the agent\u2019s access to specific relevant websites, and sensitive actions require explicit user sign-off.

For enterprise IT administrators, Google is packaging Auto Browse as a controlled deployment through Chrome Enterprise policies. Admins can define which sites and tasks are eligible for automation, set guardrails around data handling, and monitor AI activity through existing Chrome management consoles. Organizational prompts are not used for model training\u2014a critical assurance for enterprise customers wary of data leakage.

Beyond Auto Browse: Skills and Enterprise Security

The Chrome overhaul extends well beyond a single feature. Google also introduced Chrome Skills\u2014saved, reusable AI workflows that employees can invoke with a forward slash from the address bar. Think of them as macros for the AI age: one-click shortcuts for tasks like summarizing articles, extracting pricing data from competitor pages, or formatting research notes. A pre-built Skills library will ship alongside the ability for users to create custom workflows.

On the security front, Google rolled out Shadow IT risk detection, giving IT teams visibility into which unsanctioned AI tools and SaaS applications employees are using. Chrome Enterprise Premium, priced at six dollars per user per month, adds real-time data loss prevention controls covering copy, paste, upload, download, and print actions, along with data masking, dynamic watermarking, and zero-trust access controls. A new integration with Microsoft Information Protection will enforce Microsoft 365 sensitivity labels natively within Chrome\u2019s DLP engine, expected in the second half of 2026.

Google also expanded its security partnerships, announcing Device Bound Session Credentials with Okta to combat session hijacking, and advanced extension telemetry designed to detect anomalous agent activity\u2014a nod to the emerging risks that come with giving AI agents browser access.

Analysis: The Browser as Battleground

Google\u2019s timing is strategic. The enterprise browser market has attracted serious competition: Island, a dedicated enterprise browser startup, reached a 4.85 billion dollar valuation in March 2025 after raising 250 million dollars and signing 450 enterprise customers, including seven of the ten largest financial institutions. Palo Alto Networks acquired Talon in 2023 and integrated it into Prisma Access. Even Opera has launched agentic browsing features.

But Google holds a structural advantage none of these challengers can match: Chrome\u2019s near-ubiquitous installed base. By embedding AI agent capabilities directly into the browser enterprises already manage and deploy, Google sidesteps the adoption friction that standalone enterprise browsers face. The question is whether IT leaders will trust an AI agent operating inside the same browser employees use for everything else\u2014or whether the security surface area is simply too large.

There is also the productivity paradox to consider. Research has suggested that AI tools are not always reducing work but intensifying it, as managers raise expectations in response to perceived efficiency gains. Auto Browse may save an employee thirty minutes on expense reports only to see that time immediately reallocated to additional tasks.

What Comes Next

Auto Browse is rolling out in preview to Google AI Pro and Ultra subscribers in the United States, available on Windows, macOS, and Chromebook Plus devices. Enterprise deployment will be managed through Chrome Enterprise policies for eligible Workspace users.

Google\u2019s Gemini Enterprise business is already growing at 40 percent quarter over quarter as of the first quarter of 2026, and the company\u2019s infrastructure now processes 16 billion tokens per minute, up from 10 billion. Alphabet stock rose 2 percent on the day of the announcement, contributing to a Nasdaq record high.

The move signals that Google sees the browser\u2014not the chatbot, not the standalone agent\u2014as the natural home for AI in the workplace. Whether that bet pays off depends on whether enterprises are ready to let an AI agent browse the web on their employees\u2019 behalf, even with a human standing by to approve the final click.